There is something weird going on - I'm using a G5 mac (64-bit
kernel, but fully 32-bit userspace at the moment, using linux32 to
force uname to report 'ppc') trying to build trunk from a few days
ago.

 My first attempt was from CLFS-1.0.0, but I couldn't get bash in
the temporary system to compile.  Worked around it by starting the
build on x86_64.  Now that I've got the new system running and part
of my desktop  built, I'm retesting to see if it was a one-off, but
the problem is still there.

gcc -c  -DHAVE_CONFIG_H -DSHELL  -I. -I..  -I.. -I../include
-I../lib -I.    -g -DCROSS_COMPILING mkbuiltins.c
gcc -rdynamic -g -O2 -o mkbuiltins mkbuiltins.o -ldl
mkbuiltins.o: In function `main':
/mnt/lfs/building/bash-3.2/builtins/mkbuiltins.c:247: undefined
reference to `x__nldbl_fprintf'
/mnt/lfs/building/bash-3.2/builtins/mkbuiltins.c:262: undefined
reference to `x__nldbl_sprintf'

 And lots more, for each reference to fprintf or sprintf in
mkbuiltins.c.  Searching around, x__nldbl_fprintf only occurs in
mkbuiltins.o, although  __nldbl_fprintf is in libiberty.  I've tried
bash-3.2 without a patch, with each of fixes-{1..4}, and with fixes-4
plus a newer upstream (009) patch - all fail the same way.

 Maybe it's a side effect of the 64/32 kernel/userspace split,
although that used to all work fine.  Building on this box as ppc64
from ppc64 worked fine a few weeks ago (with the fixes-4 patch and
the same versions of the toolchain packages).  I can't make head or
tail of where this is coming from - any suggestions gratefully
received.

ĸen
-- 
das eine Mal als Tragödie, das andere Mal als Farce
_______________________________________________
Clfs-dev mailing list
Clfs-dev@lists.cross-lfs.org
http://lists.cross-lfs.org/cgi-bin/mailman/listinfo/clfs-dev

Reply via email to